Phenotypic and Genetic Heterogeneity of NPM1 Mutated AML at Diagnosis and during Course of Disease

Introduction:

AML with NPM1 mutation represents the largest genetic AML entity, detected in around 30% of adults with newly diagnosed disease. Despite overall favorable outcome it is highly heterogeneous with respect to mutational pattern, clonal hierarchy, blast immunophenotypes (IP) and gene expression profiles. Patients therefore may benefit from improved sub-classification at diagnosis. Flow cytometry is frequently applied and essential at diagnosis for lineage discrimination and marker-based blast assessment.

Aim:

To subclassify NPM1 mutated (mut) AML patients at diagnosis based on IP and to correlate IP with mutation profile. To compare flow cytometry based MRD analysis with NPM1 based MRD status and the presence of additional mutations during course of disease. To distinguish clonal hematopoiesis (CH) from leukemic co-mutations.

Methods:

We retrospectively analyzed 80 NPM1mut AML patients (median age: 59 years [16-92]; female: 46%; 86% intensive chemotherapy, 14% venetoclax; at least 1 follow-up sample analyzed by flow cytometry and molecular genetics) at diagnosis and sub-classified them into 5 groups (gr) based on the IP: primitive gr1, CD34-HLA-DR+CD117+/(+); “APL-like” gr2, CD34-HLA-DR-CD117+/(+); monocytic gr3, CD34-HLA-DR+CD117-CD64+CD11b+CD13(+)/-CD14+/-CD15+CD56+/-; myelomonocytic gr4, composition of gr1 or 2 with gr3, consisting of two blast populations; CD34 positive gr5: CD34+HLA-DR+. Molecular genetics (panel by NGS and RNA-based qPCR for NPM1) were compared to flow cytometry results at diagnosis and regarding MRD results (Chi-square test for MRD status comparison and Pearson correlation coefficient r for quantitative comparison). Variant allele frequency (VAF) differences of ≥5% were considered indicative of hierarchical clones.

Results:

At diagnosis cases were classified by IP as gr1: 36% (29/80), gr2: 29% (23/80), gr3: 9% (7/80), gr4: 21% (17/80) and gr5: 5% (4/80). Mutational and hierarchical profiles at diagnosis confirmed overall heterogeneity: NPM1mut did not possess the highest VAF, i.e. was not in the largest clone in 66% of gr1 (19/29), 61% gr2 (14/23), 71% gr3 (5/7), 59% gr4 (10/17) and 75% gr5 (3/4) samples. A trend for DNMT3A co-mutations was predominantly found in gr1 (41%, 12/29; 11/12 within largest clone, NPM1mut in a subclone), gr3 (43%, 3/7; 3/3 within largest clone, NPM1mut in a subclone) and gr4 (59%, 10/17; 9/10 within largest clone: 4/9 together with NPM1mut in largest clone). Gr2 most frequently harbored IDH2 co-mutations (39%, 9/23; 8/9 within largest clone: 1/8 together with NPM1mut in largest clone). Regarding samples per IP with DNMT3Amut within largest clone gr1 showed in 8/11, gr3 in 3/3, and gr4 in 5/9 samples additional other mutations within the largest clone, affected genes comprised FLT3, TET2, IDH1, NRAS, CEBPA, SRSF2. In gr2 6/8 samples had additional mutations within the largest clone together with IDH2mut.

During disease course NPM1 negative cases revealed persistent DNMT3Amut in 50% of gr1 (6/12), 67% of gr3 (2/3), 50% of gr4 (5/10) and persistent IDH2mut in 44% of gr2 (4/9), most likely reflecting pre-leukemic CH clones, whereas the majority of remaining samples were consistent with leukemic co-mutations.

Comparison of MRD status, between flow cytometry and NPM1 (qPCR) during disease course showed higher concordance in gr1, 2, 4 and 5 (57%; 67%; 51%, 60% concordance) as compared to gr3 (37%: gr1/2 vs gr3: both p<0.03). Interestingly, in gr4 the percentage of CD117 positive compartment of AML revealed poor correlation with NPM1mut (qPCR) in patients having NPM1mut within the largest clone (r=0.54), however strong correlation with DNMT3Amut if NPM1mut was detected in a sub-clone (r=0.97). In contrast, percentage of CD117 negative compartment of AML in gr4 revealed strong correlations with both NPM1 (r=0.92) and DNMT3A mut (r=0.91) if NPM1 or NPM1 and DNMT3A mut were within the largest clone. In addition, in gr2 percentage of AML cells by flow cytometry were comparable to both NPM1 (r=0.83) and IDH2 (r=0.91) mut if within the highest clone, respectively.

Conclusion:

Co-mutations in genes such as DNMT3A and IDH2 occur frequently and mostly represent the largest clone of NPM1mut AML, however, patterns differ between different IP subtypes. Phenotypic and genetic MRD analyses complement each other, further evaluation of clinical significance in subgroups will lead to a better interpretation of MRD results.

Haferlach:Abbvie: Consultancy, Honoraria.